Allowing multiple dimensions in sequence_dim

[chiaral]

Right now my understanding is that only one dimension can be used/created to concatenate files.

Example:

recipe = NetCDFtoZarrSequentialRecipe(
    input_urls=input_urls,
    sequence_dim="time",
)

Forecast data - however - they have multiple "concatenated" dimensions, or at least 2: Start time, ensemble members

As of now in this issue I am concatenating the files along Start time. Something like the following:

recipe = NetCDFtoZarrSequentialRecipe(
    input_urls=input_urls,
    sequence_dim=["time", "ensemble"]
)

would be required, with some type of nested processing of the data.

To be honest, I don't think this is a super urgent issue. We can easily create a file for each ensemble member - and then the user can easily xr.merge or xr.combine_by_coords the various ensemble members. It might actually be a good compromise to handle separately the ensemble members that have different time resolution. We simply generate one file for each ensemble member. Still, I wanted to report this need.